Investigation of Transpiration Cooling Effectiveness for Air- Breathing Hypersonic Vehicles

Size: px

Start display at page:

Download "Investigation of Transpiration Cooling Effectiveness for Air- Breathing Hypersonic Vehicles"

Roy Stafford
5 years ago
Views:

1 17th AIAA International Space Planes and Hypersonic Systems and Technologies Conference April 2011, San Francisco, California AIAA Investigation of Transpiration Cooling Effectiveness for Air- Breathing Hypersonic Vehicles S. Gulli 1, L. Maddalena 2 The University of Texas at Arlington, Arlington, TX S. Hosder 3 Missouri University of Science and Technology, Rolla, MO, The thermal management of air-breathing vehicles presents formidable challenges. The high dynamic pressure flight trajectories, the necessity of reducing the aerodynamic drag, the extended flight duration time and the need for a reusable Thermal Protection System (TPS) are stringent requirements. The work presented in this paper is focused on transpiration cooling and investigates the effects of fluid injection into the hypersonic laminar boundary layers. In particular, a simulation model, which is composed of a coupled solution of Self-Similar Method (SSM) and a Difference-Differential Method (DDM), is introduced to study the transpiration cooling along a flat plate. The reduced order code is intended to assess the boundary layer characteristics and will serve as a research tool for the design and analysis of future experimental investigations in the UTA s 2MW arc-heated facility that has been modified and is currently in use for TPS studies. The DDM solves a system of coupled algebraic and Partial Differential Equations (PDE) for the case of Pr=1 and Le=1. Self-Similar solutions are considered in order to compare the code results for the case without transpiration. Nomenclature BC = Boundary condition = Partial Derivative BL = Boundary layer = Incremental ratio DDM = Difference Differential Method = Density L/D = Aerodynamic Efficiency = Viscosity NS = Species Number = Longitudinal transformed coordinate PDE = Partial Differential Equation = Transversal transformed coordinate SSM = Self-Similar Method = Rate of species production C = Chapman-Rubesin parameter = Concentration of air Subscripts = Concentration of coolant e = External L = Flat plat length CF = Coolant Fluid P = Pressure w = Wall T = Temperature m = Matrix TPS = Thermal Protection System i = Initial = Longitudinal velocity V = Transversal velocity Le = Lewis number Pr = Prandtl number f, g = Service functions for self-similar transformations h = Static enthalpy 1 Visiting Scholar, Aerospace Engineering, 500 W. First Street, Arlington, TX 76019, AIAA Member. 2 Assistant Professor, Aerospace Engineering, 500 W. First Street, Arlington, TX 76019, AIAA Senior Member. 3 Assistant Professor, Aerospace Engineering, 400 West 13 th Street, Rolla, MO 65409, AIAA Senior Member. 1 Copyright 2011 by S. Gulli, L. Maddalena, S. Hosder. Published by the, Inc., with permission.

2 I. Introduction Reusable thermal protection systems are one of the key technologies that have to be improved in the design of hypersonic vehicles. When considering hypersonic air-breathing systems such as a two-stage-to-orbit (TSTO) access to space vehicle or a hypersonic cruise vehicle for global engagement, new challenges have to be overcome in order to enable this technology. Among these, the high dynamic pressure flight trajectories, the extended flight duration time and the necessity of reducing the aerodynamic drag can be listed. As an example for the extreme thermal loads encountered by these vehicles, the heat flux to a leading edge of a 2 inches radius of curvature on an hypersonic vehicle flying at Mach 10 at 30 km is approximately 200 W/cm 2. If the radius of curvature of the leading edge is reduced to 0.25 inches to improve the aerodynamic efficiency, the heat flux, becomes approximately 500W/cm 2. When such extremely high thermal loads are encountered, the maximum surface temperature for which the specific strength and the environmental durability of the material are preserved can be greatly exceeded and active cooling systems should be considered for the thermal management. This paper concerns with a preliminary study of transpiration cooling and its effect on the boundary layer structure, the blockage effect and the consequent reduction of heat loads. Boundary layer simulations are fundamental for hypersonic problems where viscous effects are of primary importance. The overall aerodynamic performance of every flying vehicle is strongly dependent on the effects near the wall. In hypersonic flows, the viscous effects near the wall have greater importance from the point of view of thermal loads (i.e., heat flux and temperature distributions) and aerodynamic performance (i..e, L/D). For this reason it is fundamental to understand the physics that characterize the boundary layer and simulate its behavior for different surface parameters such as the type of material, surface manufacturing, surface coating, wall geometry etc. The work presented in this paper is focused on transpiration cooling and investigates the effects of fluid injection into the hypersonic laminar boundary layers. In particular, a reduced simulation model is introduced to study the transpiration cooling along a flat plate. The reduced order code is intended to assess the boundary layer characteristics and will serve as a research tool for the design and analysis of future experimental investigations in the UTA s 2MW arc-heated facility that has been modified and is currently in use for TPS studies. II. Review on Boundary layer Solution Procedures In the last 80 years many solution methods have been proposed to solve the boundary layer problem after the first theoretical study by Prandtl (1904) 1. Two main approaches are possible to solve the boundary layer equations: The integral method and the differential method. The integral methods solve the Navier-Stokes equations in the integral form applying the conservation principles to a control volume that is finite in the transverse direction (where the variation of thermodynamic properties is more important), and differential in the streamwise direction. The differential methods apply the conservation principles to a differential fluid s volume. The integral methods are computationally less expensive compared to the differential methods, but they are not suitable for a precise estimation of quantities such as C or q since with the integration all the y-history of the thermodynamic properties is lost and only the boundary conditions are taken into account 4. In the integral method, in order to extract velocity or temperature profiles along the y direction, it is necessary to consider polynomial expressions for the velocity (Pohlhausen method) 3. Otherwise, one can consider non dimensional functions to simplify the problem and solve the integral equations (Thwaites-Walz method) 3. This type of transformation cannot extract the variance of thermodynamic properties along the y-direction. The differential methods, although mathematically more complex, can describe in detail the flow structure in the boundary layer. They permit the computation of the velocity and temperature profiles allowing the subsequent calculation of the skin friction and the heat transfer on the wall. Considering the differential method approach, it is possible to adopt, for certain cases, Self Similar Methods (SSMs) and, in general, Difference Differential Methods (DDMs) for the solution of the governing equations. The SSM transforms the Cartesian space (x; y) to another (ξ; η) space by means of different coordinate transformations (i.e. Illingworth-Levy or Levy-Lees). Self-Similar solutions have been derived for the flat plate geometry and incompressible flow (Blasius equation) 1,2,3,4. For compressible boundary layer flows, with the introduction of Chapman-Rubesin coefficient 3 : 1.1 The x-momentum and energy equations in the transformed space become 1 : 2

3 Cf ff 2ξ C Pr g fg C U f 2 h These equations are valid for any geometry and with the assumption that the pressure remains constant in the transverse direction in the boundary layer. After calculating the solution in terms of f & another operation is then needed to re-transform the (ξ; η) space to (x; y) space 1. The SSM has the main advantage to transform the PDEs to ODEs that can be easily solved. On the other hand, the DDM solves the PDEs given in (1.2) directly by discretizing the boundary layer equations on a computational mesh. It then solves the discretized equations using a combination of forward, upward and central finite difference approximations for the partial derivatives 9. III. Mathematical Model The capability to compute the thermodynamic properties at all points inside the boundary layer (BL) is fundamental in resolving the near-wall the hypersonic flow field. For this reason, as reported in Section II, the solution to the BL problem has been approached with the differential method. The general equations for the hypersonic BL with the relative assumptions are reported below: - 2-D Flow Ideal gas - 0 (negligible radiation) - 0 (non-reacting flow) , The system of equations (1.3) is valid for a thin BL at high Reynolds numbers. It is worthy to remark the fact that the third equation (pressure variation through the BL) is no longer valid for very large Mach numbers. In the current model, in addition to the assumptions above, the following simplifications are imposed on the model: - 0 (Flat Plate Geometry) With these simplifications, the governing systems of equations become: 3

4 0 0 μ ,2. For the special case of Pr=1 and Le=1, the energy equation and the species equation have been analyzed by Crocco 7,8 which lead to a simple solution. With this, the set of equations for two species (with air transpiration) are reduced to: The transpiration cooling is taken into account with the boundary condition, which specifies the transverse velocity at the wall ( ).The system (1.5) is composed of two PDEs and six algebraic equations. Note that this mathematical model solves simultaneously algebraic equations and PDEs. In the previous studies 7,8,4, to obtain a rapid estimation of BL properties with transpiration, the algebraic energy and species equations were implemented into SSM with particular values for the injection parameter to preserve the self-similarity. The complete equations (1.3) have been solved with high-fidelity CFD techniques (i.e., full Navier-Stokes calculations). With the current mathematical model (1.5), we are able to estimate the effects of transpiration without constraints on the injection velocity and without using computationally expensive CFD models In the current model, it is possible to relate the wall temperature with the injected mass flow rate. The heat balance is imposed (Figure 1) considering 1-D analysis, neglecting, for this particular case, the radiation heat and assuming that the wall material is thermally isotropic 2, 5, 12. 4

5 Figure 1. Heat balance at the porous wall The changes of the thermodynamic properties of the coolant into the material are based on the convective heat load on the surface. From the heat balance, given the initial temperature of the coolant fluid, is possible to find how the coolant and matrix temperature varies through the material. This variation will be dependent on the mass flow rate, the coolant fluid properties and material characteristics. Given the assigned wall temperature, imposed by the material characteristics of the thermal protection system, it is possible to solve the BL problem and determine the wall heat transfer. The coupling of the mathematical models for the BL and for the material allows studying how the wall heat transfer is influenced by the injection parameters, material characteristics and by the type of coolant fluid. IV. Numerical Solution It is possible solve the differential transport equations (1.5) by the SSM, the DDM, or a coupling of both. The SSM is preferable to the DDM because it transforms the PDEs to ODEs that can be solved easily. The SSM is not considered for our transpiration problem since a streamline function has to be introduced to transform the equations in (; ) plane to the (; ) plane. The streamline function assumes a well-known form when mass addiction is not considered 1,11. When transpiration is considered, an additional function, of which we do not know an explicit expression, has to be modeled. In literature, all transpiration problems were treated with the SSM considering a binding expression for the transpiration parameter that is ~ (12). This condition on transverse velocity, which is not useful for general transpiration cases, leads to a self-similar solution obtained from the resulting system of ODEs (Hartnett and Eckert solution 2,6,12 ). In this paper, a general purpose code that utilizes the DDM method has been implemented. To start the solution with this method, an initial solution is needed in addition to the boundary conditions at the wall and at the outer edge of the BL. For the geometry we consider in this paper (flatplate), the initial solution is taken as the self-similar solution for compressible flow on a flat plate. In this fashion, we are implementing a coupled SSM and DDM. For different geometries such as the axisymmetric bodies, it will be possible to use only the DDM. This is motivated by the fact that the initial solution can be extracted from the stagnation point solution. For this reason, the initial part of the flat plate is modeled without transpiration. The system of equations to obtain the starting solution are derived from Eqn 1.2 with the assumption of a flat-plate geometry: Cf ff 0 C Pr g fg C U f 0 h 1.6 in which the Sutherland s law is used to calculate the dynamic viscosity in Chapman-Rubesin parameter 11 / 1.7 5

6 where the S = 111K for air. After solving the compressible BL without transpiration for the initial portion of the flat plate, the system (1.5) is implemented using an implicit method to avoid any instabilities in the solution. A mix of forward and central finite differences 9 has been used. V. Validation and Mesh Sensitivity Surface heat transfer comparisons obtained with the current method to the results obtained with the CFD code GASP are reported in Figure 2. The GASP code was used to obtain the numerical solution of the full Navier-Stokes equations for the laminar compressible flow over a flat plate with transpiration cooling, which was modeled by imposing a constant transverse velocity at the wall. The simulations have been performed for three cases with 0 / (no transpiration); 0.1 /; and 0.3 / activating the air transpiration at 0.3 (measured from the leading-edge of the flat plate where x=0.0 m). For all the cases, the freestream Mach number was taken as 5.0, the freestream temperature as 300K, and the freestream density as 1.23 kg/m 3. In addition, the wall temperature was kept at a constant value of 1000K for all the cases. Figure 2. Wall heat fluxes comparison a. b. Figure 3. Comparison of the Boundary layer profiles a. GASP code b. UTA University code The maximum difference in heat transfer between the two approaches is about 10% near the zone where the transpiration starts. The maximum error on the BL thickness distribution (shown in Figure 3) is about 20%. This difference may be due to two reasons: (1) The UTA code uses constant thermo-physical properties (,). (2) The GASP code can capture the shock at the leading edge, which is evident from Figure 3. For x = 0 m the UTA code shows a zero BL thickness while the GASP results indicate a small but non-zero BL thickness at the same location. The future work will also target to explain the exact reason(s) of this difference. Overall the comparisons show the reasonable agreement between two approaches, which indicates the potential of the method introduced in this paper. A grid-sensitivity analysis was performed using two mesh levels and comparing the results to the SSM results for the no-transpiration case. 6

7 a. b. c. Figure 4. Mesh sensitivity for DDM and comparison with SSM results a points ; b points ; c. Self-similarity solution with 1M points Figure 4 c, shows that the results derived from the coupled DDM+SSM without transpiration follow the Self-Similar solution (Figure 4.a.). At the end of the flat plate (x=1.0 m) the BL thickness of the SSM+DDM solution ( ) with grid points is 0.04% higher compared to the SSM solution (, ). The fluctuations on the solution shown in Figure 5.b. are due to the rounding and truncation errors of the numerical method. VI. Preliminary Results After the code was compared with the self-similar solution for the case of no-transpiration and the grid convergence was assessed, the transpiration was enabled. The transpiration effects were investigated varying the transversal velocity at the wall. The main input parameters (the free stream conditions and the wall temperature) were the same as the ones used in the validation study. a. b. c. d. Figure 5. Effects of suction/injection of air into the hypersonic BL a.. ; b.. ; c.. ; d.. 7

In Figure 5, the blue solutions represent the cinematic BL while the red curves represent the thermal BL. Figure 5 a. represents the case of constant suction velocity = -0.

1 1.025 0.98 1.7 This ratio is obtained as 0.95 from the simulation results. The BL thickness at x=1.0 m for the case with 0.

3 m/s, the longitudinal velocity U becomes negative and the method becomes unstable (Figure 6). This can be observed beyond the location, where we see the laminar BL separation (i.

8 In Figure 5, the blue solutions represent the cinematic BL while the red curves represent the thermal BL. Figure 5 a. represents the case of constant suction velocity = -0.3 m/s while Figures 5b, 5c, and 5d report, respectively, the case of constant injection velocity =0.1, 0.2 and 0.3 m/s. The difference between the cinematic and thermal BL remain constant for each case because it depends only on the Prandtl number, which is equal to unity in the current study This ratio is obtained as 0.95 from the simulation results. The BL thickness at x=1.0 m for the case with 0.3 m/s transpiration wall velocity is about 80% more compared to the case with no-transpiration. If the transverse velocity is increased beyond 0.3 m/s, the longitudinal velocity U becomes negative and the method becomes unstable (Figure 6). This can be observed beyond the location, where we see the laminar BL separation (i.e, where the velocity gradient in the transverse direction become zero at the wall) near x=0.95m for a transverse velocity of 4 m/s (Figure 6). Figure 6. Longitudinal velocity for =0.4m/s The effects of transpiration on the thermodynamic properties are clear looking at the temperature profiles (Figure 7) and the heat transfer at the wall (Figure 8). a. b. c. Figure 7. Temperature profiles of the thermal BL for different air injection velocity a../; b../; c../ At the same x-station, the increase in transverse (transpiration) velocity increases the thermal boundary layer thickness and gradually reduces the temperature gradient at the wall. A zoomed view on the wall (Figure 8) makes this result more clear. 8

a. b. Figure 8. Temperature profiles at the wall a. =0.

3 m/s This trend affects directly the heat transfer at the wall that is proportional to the temperature

Heat transfer results along the flat plate At the end of the plate, the heat flux to the wall is 34 for the

9 a. b. Figure 8. Temperature profiles at the wall a. =0.1 m/s; b. =0.3 m/s This trend affects directly the heat transfer at the wall that is proportional to the temperature gradient at the wall (Figure 9). Figure 9. Heat transfer results along the flat plate At the end of the plate, the heat flux to the wall is 34 for the non-transpiration case ( 0/), whereas it drops to a value of 1.3 for the ( 0.3 / case. The skin friction, as the heat transfer at the wall, decreases when the transpiration velocity increases and the gradients within the boundary layer are reduced (Figure 10). Figure D Thermal BL profiles for different injection velocities 9

10 VII. Conclusion The work presented in this paper is focused on transpiration cooling and investigates the effects of fluid injection into the hypersonic laminar boundary layers. In particular, a reduced simulation model is introduced to study the transpiration cooling along a flat plate in hypersonic flow, which is computationally less expensive compared to the high-fidelity CFD techniques. In the development of the method, a differential approach has been preferred over the integral approach to acquire more detailed information on the thermodynamic quantities inside the boundary layer. The mathematical model couples the solution of continuity and x-momentum in partial differential equation form with the energy and species equations in the algebraic form obtained for Pr=1.0 and Le=1.0. The mass injection is taken into account with a boundary condition, which specifies the transverse velocity at the wall. The transpiration cooling problem is solved by the Difference Differential method (DDM) without utilizing the Self-Similar Solution method (SSM), which is known to give a solution only in the case of ~. The SSM is used only as the starting point for the DDM. The numerical results obtained for three transpiration cooling cases with different injection velocities show the importance of having the capability to predict the variation in various boundary layer quantities, particularly the wall heat transfer and skin-friction, with different injection parameters. (velocity and mass flow rate), the type of injected fluid, and the wall material characteristics (porosity, permeability etc.). The comparisons between the results of the developed method and the high-fidelity CFD code GASP have shown that the method is ready for the coarse prediction of the wall heat flux and the skin-friction distributions in hypersonic laminar boundary layer flows with transpiration cooling. VIII. Future Work Future work will be focused on the improvement of the code by removing the 1 and 1 condition. This will allow the consideration of any type of injected fluid and a variable in the boundary layer. The future adoption of variable thermo-physical properties through the boundary layer is also expected to improve the accuracy of the method. For generality, a coordinate transformation scheme will be implemented to apply the developed method to any geometry. The boundary layer separation criteria and turbulent boundary layers will be also considered. The code will be used to guide the experimental component of the studies on transpiration cooling for hypersonic vehicles. Similarly, improvements to the code will be supported by the experimental investigations as well. References 1 Anderson, John D., Jr., Hypersonic and High Temperature Gas Dynamics, Mc Graw Hill (1989), pp Burmeister, Convective Heat Transfer, Wiley-Interscience (1983), pp Frank, M. White, Viscous Fluid Flow, Mc Graw Hill (1974), pp Joseph, A. Schetz, Boundary Layer Analysis, Prentice Hall (1993), pp ; pp David, E. Glass, Numerical Analysis of Convection/Transpiration Cooling, (1999) NASA TM A., L. Laganelli, R., P. Fogaroli, A., Martellucci, The Effects of Mass Transfer and Angle of Attack on Hypersonic Turbolent Boundary Layer Characteristics, April 1975, AFFDL TR-75-35, Appendix A. 7 B. W. Van Oudfheusden, Compressibility Effects on the Extended Crocco Relation and the Thermal Recovery Factor in Laminar Boundary Layer Flow, FEDSM , ASME May 29 June 1, 2001, New Orleands, Louisiana. 8 E., R. Van Driest, Investigation of Laminar Boundary Layer in Compressible Fluid Using the Crocco Method, (1952), NACA-TN Md Tanvir, R. Faisal, A. K. M. S. Islam, Primitive Variable Approach to Calculate Separation Point in Laminar Boundary Layer Along a Flat Plate, ARPN Journal of Engineering and Applied Science, Vol.1-No.3, October John, H. Lienhard IV, John, H. Lienhard V, A Heat Transfer Textbook, Phlogiston Press (2006), pp Y. Cho, A., Aessopos, Similarity Transformation Methods in the Analysis of the Two Dimensional Steady Compressible Laminar Boundary Layer, Term Paper, 2.26 Compressible Fluid Dynamics, Massachusetts Institute of Technology, Spring Richard A. Thompson and Peter A. Gnoffo, Implementation of Blowing Boundary Condition in the LAURA Code AIAA Paper TH Aerospace Sciences Meeting & Exhibit. Reno, NV: 7 10 January E.R.G. Eckert, John N.B. Livingood, Comparison of Effectiveness of Convection-, Transpiration-, and Film-Cooling Methods with Air as Coolant, (1954) NACA Report

A Non-Intrusive Polynomial Chaos Method For Uncertainty Propagation in CFD Simulations

A Non-Intrusive Polynomial Chaos Method For Uncertainty Propagation in CFD Simulations An Extended Abstract submitted for the 44th AIAA Aerospace Sciences Meeting and Exhibit, Reno, Nevada January 26 Preferred Session Topic: Uncertainty quantification and stochastic methods for CFD A Non-Intrusive